When an airborne platform on which Synthetic Aperture Radar (SAR) mounted on flies very complicatedly over subaperture time, traditional quadratic phase error can not depict the errors introduced by the platform motions, and the phase errors need be modeled as polynomial errors with higher order. A new algorithm to extract polynomial errors with higher order from SAR echoes, named as Time Phase Adjustment by Contrast Enhancement (TPACE), is proposed in this paper. The TPACE takes the contrast function as object function and the polynomial coefficients of phase errors as variables to be estimated, directly extracting those coefficients by optimization. The gradient of the contrast function with respect to the variables, which is the key point to implement TPACE algorithm, is derived detailed in the paper. Compared to those existed algorithms that can extract polynomial phase errors, the computation load of TPACE is much less and the TPACE is very useful for real time processing. The results processing real ultra-wide band SAR echoes show the method can extract effectively polynomial phase errors from SAR echoes.